MapReduce was the only use case on which the previous versions of Hadoop were developed. We know that MapReduce is mainly used for the efficient and effective processing of big data. It is used to process a graph and millions of its nodes and edges. Going forward with technology, to cater for the requirements of data location availability, fault tolerant systems, and application priorities, YARN built support for everything from a simple shell script application to a complex MapReduce application.

For the data location availability, MapReducer's ApplicationMaster has to find out the data block locations and allocations of containers to process these blocks accordingly. Fault tolerant system means the ability to handle failed tasks and act on them accordingly, such as to handle failed map and reduce tasks and rerun them with other containers if needed. Priorities are assigned to each application in the queue; the logic to handle complex intra-application priorities...